Diagnostic performance of RMI I–IV for pelvic/adnexal masses after hysterectomy: a retrospective single-center study

To evaluate and compare the diagnostic performance of four Risk of Malignancy Index (RMI) models (I–IV) in differentiating benign from malignant pelvic masses newly developed in post-hysterectomy patients. This 10-year retrospective single-center study enrolled 171 patients with newly developed pelvic/adnexal masses after hysterectomy who received surgical resection and pathological diagnosis. Baseline clinical and imaging parameters—including age, CA125 level, tumor size, ultrasound features, and menopausal status—were collected. Receiver operating characteristic (ROC) curves and DeLong's test were used to compare the area under the curve (AUC) of RMI I–IV, while the Youden index was used to determine the optimal cutoff for each model. Diagnostic sensitivity, specificity, and accuracy were calculated using both traditional published thresholds and cohort-specific optimal cutoffs. Numerically, RMI I exhibited the highest specificity (98.6%) and the largest AUC value (0.973, 95% CI: 0.942–1.000) among the four evaluated indices. Overall, all RMI models demonstrated good diagnostic performance. At a cutoff of ≥ 60, RMI I achieved a sensitivity of 96.7% and a specificity of 91.5%. In pairwise DeLong's tests, the AUC of RMI I was significantly higher than that of RMI II and RMI IV ( p = 0.009 and 0.013, respectively), while its advantage over RMI III did not reach statistical significance ( p = 0.057). Importantly, however, none of the RMI models yielded satisfactory sensitivity for detecting FIGO stage I–II epithelial ovarian cancers. Among the four models, RMI I demonstrated superior preoperative risk stratification value for post-hysterectomy pelvic masses. Nevertheless, the present study is subject to several limitations, including its single-center retrospective design, inherent surgical selection bias, and a relatively small sample with only 30 malignant cases. Moreover, the optimal thresholds derived from this cohort are exploratory and should not be generalized to routine clinical practice without external multi-center validation. As such, RMI should be regarded as an auxiliary assessment tool rather than the sole basis for clinical decision-making.